In April 2003, the Swedish philosopher Nick Bostrom, then a researcher at the University of Oxford, published a 13-page paper in The Philosophical Quarterly titled “Are You Living in a Computer Simulation?” The paper presented a rigorous argument that has, in the more than two decades since, become one of the most widely discussed pieces of contemporary academic philosophy. Bostrom argued that anyone considering whether they live in genuine biological reality or inside a computer simulation should take the simulation possibility much more seriously than common sense suggests. His argument did not, contrary to popular shorthand, claim to prove that we are inside a simulation. It proved a trilemma — that one of three specific propositions must be true — and the simulation conclusion is only one of the three.
The argument has been promoted by tech entrepreneurs (Elon Musk has cited it repeatedly), explored by philosophers (David Chalmers wrote an entire 2022 book, Reality+, building on Bostrom’s framework), and translated into popular media to the point that “we might be in a simulation” has entered everyday conversation in a way few philosophical theses ever achieve. The popular framing is generally that Bostrom argued we are probably living in a simulation. Bostrom’s actual argument is more careful, and the conditional structure matters for how seriously the claim should be taken.
The trilemma
According to Bostrom’s original 2003 paper, the argument runs as follows. Begin by assuming that human-level civilizations like our own can, with enough technological development, reach what Bostrom calls a “posthuman” stage — one capable of constructing computer simulations detailed enough to contain conscious simulated minds. Such simulations are called “ancestor-simulations” because the most likely thing a posthuman civilization would simulate, given the cognitive and historical interest of doing so, is its own evolutionary and historical past — including the conscious experiences of the people who once lived in it.
From this starting point, Bostrom argued, at least one of the following three propositions must be true:
The first is that the fraction of human-level civilizations that reach a posthuman stage is very close to zero. In other words, civilizations like ours almost always go extinct, regress, or otherwise fail to develop the computational capacity to run ancestor-simulations before something destroys them. The second is that the fraction of posthuman civilizations that actually choose to run a significant number of ancestor-simulations is very close to zero. In other words, advanced civilizations might exist but, for ethical, practical, or aesthetic reasons, decline to simulate their ancestors. The third is that we are almost certainly living in an ancestor-simulation ourselves.
The logic that produces the trilemma is straightforward. If posthuman civilizations exist (negating proposition 1) and they do run ancestor-simulations (negating proposition 2), then the total number of simulated minds throughout history vastly exceeds the number of biological minds. The computational resources available to a posthuman civilization would be sufficient to run billions or trillions of detailed simulations, each containing billions of simulated conscious beings. The biological minds in any single civilization are a finite, comparatively tiny set. By simple counting, any given conscious experience — yours, mine, the experience of reading this article — is statistically much more likely to be one of the many simulated experiences than one of the few biological ones.
The bland indifference principle
The mathematical bridge between “most minds in this kind of universe are simulated” and “I am probably simulated” relies on what Bostrom calls the “bland indifference principle.” The principle states that, if you have no information that distinguishes you from other conscious beings in your epistemic position, you should assign yourself an equal probability of being any of them. According to the Wiley Online Library reference for Bostrom’s paper in The Philosophical Quarterly, this principle is roughly analogous to the reasoning used in standard anthropic-principle arguments: if you don’t know which of a large group you are, assume you’re equally likely to be any of them.
The principle has been disputed in the philosophical literature. Brian Weatherson, in a 2003 reply also published in The Philosophical Quarterly, argued that Bostrom’s indifference principle was either too strong (implying conclusions that seem clearly wrong in other domains) or insufficiently general. Bostrom replied in 2005 with a defence of the principle’s specific application to the simulation case. The debate continues. What is not in dispute is that some version of the principle is needed for the move from “most minds are simulated” to “you are probably simulated” to go through. If the indifference principle fails, the simulation conclusion does not follow even when propositions 1 and 2 are rejected.
What this changes about the simulation question
The most important consequence of the trilemma structure is that anyone who wants to argue that we are not in a simulation must commit to one of the first two propositions. There are not three free-standing options to choose between. The propositions are linked. Believing that we are in genuine biological reality requires believing either that civilizations like ours never reach posthuman capability (proposition 1), or that posthuman civilizations exist but choose not to simulate (proposition 2). According to a PBS Nova explainer on the simulation argument, this is what has made Bostrom’s framing so influential: it converts what seemed like a fanciful science-fiction speculation into a structured choice between specific philosophical and empirical claims.
The first proposition is essentially a statement about existential risk. If human-level civilizations almost always destroy themselves, decline, or fail to expand technologically before reaching posthuman capacity, then the simulation conclusion fails because there are no posthumans to do the simulating. This connects the simulation argument to the broader study of existential risks — nuclear war, climate change, hostile artificial intelligence, biotechnology accidents, asteroid impacts — that Bostrom has spent much of his subsequent career exploring at Oxford’s Future of Humanity Institute (which he founded in 2005 and directed until 2024).
The second proposition is essentially a statement about posthuman psychology and values. If advanced civilizations have ethical concerns about creating simulated minds that would themselves experience suffering — or if they simply lose interest in their own ancestral history once they have transcended it — then the simulation conclusion also fails, because the simulations would never be built. This is a contested empirical claim about what very advanced civilizations would actually do, which is difficult to answer because no such civilizations are known to exist.
What Bostrom himself believes
Bostrom has been careful, in subsequent writings and interviews, not to claim personal certainty about which branch of the trilemma is correct. According to a 2024 review of the simulation argument’s arithmetic by independent researcher Howard Rudd, Bostrom has typically estimated his personal credences in the three propositions as roughly equal — perhaps one-third each — with the result that he assigns roughly a one-in-three probability to actually being in a simulation. This is much higher than common-sense intuition would suggest, but considerably weaker than the popular shorthand “Bostrom thinks we’re in a simulation.”
Several refinements and critiques of the argument have appeared in the two decades since the original paper. Marcus Arvan, a philosopher at the University of Tampa, has argued for a “peer-to-peer” alternative model in which simulations might be distributed across networks rather than run on single centralised computers. Cosmologists including George Ellis have argued that the simulation hypothesis is unfalsifiable and therefore not properly a scientific question. Other philosophers have argued that consciousness cannot be implemented on computational substrates at all, which would dissolve the argument from the outset by ruling out the possibility of ancestor-simulations containing conscious minds. The debate is now substantial enough to support a small subfield of contemporary philosophy.
The simulation argument’s enduring power is partly that it requires no exotic physics, no untestable cosmology, and no appeal to religious or supernatural categories. It uses only assumptions that most contemporary scientists already accept: that consciousness can in principle be implemented on computational substrates, that civilizations may eventually develop very powerful computers, and that ancestor-simulations would be one plausible thing to do with such capacity. Given these assumptions, Bostrom’s trilemma is mathematically airtight. The conclusion is therefore not a fringe idea. It is a serious structured argument that anyone confident we are in original biological reality has to engage with, and the engagement requires committing to a claim about either the future of civilizations or the values of posthuman minds. There is no fourth option.
